eudi.tools

Synthetic, on purpose

signing key
published in this repository
claims
one invented persona
protocol
real OpenID4VCI

The exchange is real and the credentials are well-formed — that is what makes them worth testing against. What they are not is evidence about a person, or evidence that anyone trustworthy issued them: the key is public, so anyone who reads the repository can sign the same thing.

So a verifier accepting one has shown it can read an SD-JWT VC and check a signature. It has not shown it would refuse a credential from an issuer it should not trust — for that, add the anchor below to a trust list deliberately, or try the entries here that are meant to be refused.

two certificate authorities, one per role
credentials
CN=eudi.tools test root → CN=dev.eudi.tools
belongs in a PID Provider list
attestations
CN=eudi.tools wallet provider (synthetic)
belongs in a Wallet Provider list

These are unrelated keys and neither works in the other’s place. Putting the wallet provider certificate in a PID Provider list fails as chain does not lead to a trusted root — a message about the chain, which is intact, rather than about the certificate, which is the wrong one.

A credential carries its leaf in x5c and never the root — HAIP §6.1.1 forbids it, because a chain carrying its own anchor proves nothing. So the anchor has to arrive some other way: /trust has both certificates, as files a trust list takes, with the fingerprint to check each by.

Issue yourself a credential.

Each entry produces a real openid-credential-offer:// URI over OpenID4VCI, and the credential is bound to a key your wallet proves it holds — so nobody else can present it. Four of the five are meant to be refused.

Everything here is synthetic: one invented persona, signed by a key published in this repository. The point is not the claims, it is what your verifier does with them.

On the mDL entries. The mDL entries are built to the ISO/IEC 18013-5 second-edition CD ballot resolution draft (v4, dated 2025-09-14), because that is the document this project holds and has read. The published 2021 first edition is what deployed verifiers implement, and the second edition renumbers clauses and changes substance. Where the two differ, these credentials follow the draft — so if your verifier disagrees with one, check the edition before you check your code.